Theories are created in science to explain a certain set of phenomena. Intuitively, however, we tend to increase our belief in a new theory if it not only explains the phenomena that it was specifically created to address but then is found to follow from something we had already accepted. Yet in Bayesian Probability Theory, the dominant tradition concerning the confirmation of scientific theories, old evidence does not increase the probability of new theories. How should this so-called `old evidence/new explanation` issue be resolved? Richard Jeffrey came up with one solution. Professor Wagner is extending Jeffrey's principles and, in process, gives rise to two different revision methods for generalization of this problem. The PI will investigate both the mathematical and the philosophical aspects of this phenomenon, both because of the intrinsic interest of generalizations of the old evidence/new explanation problem and in order to clarify the essence of Jeffrey's solution to the classical version of this problem.
